1/10/2017

Overview
Flocculator Design
•Analysis of hydraulic flocculators
•Ratio of maximum to average energy 
dissipation rate
•Inefficiency of energy use due to 
nonuniformity of energy dissipation rate
•The great transition at  ⁄ =5
•Flocculator Design
•Head loss, collision potential, residence time
•Geometry of a baffle space to obtain desired 
energy dissipation rate

Top View Side View
H = Water depth
W = Width of the flocculator channel L = Length of the flocculator channel
S = Space between baffles
S = Space between baffles T = Thickness of the baffles S T  B
L = Length of a flocculator channel B =Perpendicular center to center distance between baffles

Exit to the sedimentation tank entrance channel

S Minimum water level
W 1.5 S Upper baffle
S H Lower baffle

L Port from
1.5S previous channel
L

Design Considerations More Design Considerations
•The length of the flocculator channels matches the  •Even number of channels for AguaClara 
length of the sedimentation tank
•Width of the flocculation channel? design (to keep chemical dose controller 
•Minimum? ________________
Human width near stock tanks), but this may change if 
•Material limitations (polycarbonate or concrete) flocculators get smaller
•Vary to optimize flocculation  •Even or odd number of baffles depending 
efficiency (function of geometry)
•Need to determine on channel inlet and outlet conditions
•Head loss •Begin with the energy source for the 
•Residence time turbulence that creates shear that creates 
•Baffle spacing collisions: head loss for a baffle
•Number of baffles

CEE 4540: Sustainable Municipal Drinking Water Treatment

Monroe Weber-Shirk 1
 1/10/2017

Vena Contracta around a bend? Vena Contracta (VC ) Conclusions 

•Sluice gate (almost closed)* •Draw the most extreme streamline 
•0.59  through the transition and determine 
•Small hole in a tank the total change in direction
•0.62 •If the change in direction for most of 
the fluid is 90°, then the VC is 
By Lindsay Lally, Lee Hixon (Own work) [CC BY-SA 3.0

•Exit from a pipe (http://creativecommons.org/licenses/by-sa/3.0) or GFDL

(http://www.gnu.org/copyleft/fdl.html)], via Wikimedia Commons

•No Vena Contracta approximately 0.62
•If the change in direction for most of 
the fluid is 180°, then the VC is 
* Roberson, JA; Cassidy, JJ; Chaudhry, MH. Hydraulic Engineering. John Wiley. (1995) page
217. Original reference is Henry, H.R. “Diffusion of Submerged Jets.” Discussion by M.L.
approximately 0.622=0.384
Alberston, Y.B. Dai, R.A. Jensen, and Hunger Rouse, Trans. ASCE, 115, (1950)

Head Loss coefficient for a Baffle Flocculator Efficiency

Head loss in an expansion Which space between baffles is

better, considering the uniformity
of the energy dissipation rate?
e - expansion

the contraction
coefficient for a sharp
180º bend (0.622) This space with very low energy
We need to measure this in one of the new AguaClara plants! dissipation rate doesn’t contribute much ⁄ =4
⁄ = 10

Simplify flocculator design by 
Why a transition at  ⁄ of 5?
designing for high efficiency
0.4S
0.6S

•Jets expand in width at the rate of  •Efficiency will be a function of the 
approximately 1 unit in width per 10 units  variability of the energy dissipation 
forward rate 

•Expansion length is 10(0.6S) •We expect a relation of the form  S
such that efficiency is 1 when
•Expansion requires a distance of  =1 and efficiency is less than 1 
approximately 6S for higher values of
•The  ⁄ transition is related to the  •We “solve” this unknown by always 
distance required for the jet to fully  designing efficient flocculators 
with 3<H/S<6
expand

CEE 4540: Sustainable Municipal Drinking Water Treatment

Monroe Weber-Shirk 2
 1/10/2017

Prior to 2015 AguaClara used designs 
New Approach: Always efficient
that were far from the optimum
•A compact plant layout was possible for small  •Add obstacles to have  30 L/s 10 L/s
flows by using a vertical flow flocculator with a  a maximum  ⁄ ratio 
high  ⁄ ratio of between 3 and 6.
•For small plants the width of the channel was  •Flocculation efficiency 
determined by the need to construct the channel 
using humans (45 cm or more)  can be considered 
•The space between baffles was very narrow and  constant (and close to 
thus  ⁄ was very high (for low flow plants) 1)
•Small plants needed longer residence time and 
more baffles to achieve adequate flocculation 
because efficiency was reduced.

Viscous collisions or inertial 
Collision Potential
collisions
•Prior to 2016 I had assumed that the  •The target collision potential used for the 
appropriate length scale comparison was  design of AguaClara plants since about 2013 
particle separation distance and Kolmogorov  has been 37,000
length scale – thus concluded inertia was  •The actual collision potential in operating 
important AguaClara plants may be lower because the 
2016
•Particle separation distances are smaller  head loss per baffle may be lower than we 
than inner viscous length scale assumed
•Collisions in turbulent flocculators are 
dominated by viscosity (fluid shear, not 
turbulent eddies)* * Edge of knowledge

Energy use (head loss) in flocculation 
The Influence of  ̅ or GMax
controls velocity gradient
•Head loss  ̅ or  ̅ determines the head loss through the 
• The value of 
•High head loss results in a taller  flocculator
building for the water treatment plant • Maximum size of the flocs is controlled by 
•High head loss means higher velocities  • ̅ or  ̅ (assuming shear limits attachment) Not yet
and that reduces settling of flocs in the  • GMax or Max (assuming floc break up controls max size) known
flocculator
• Max = 10 mW/kg (GMax = 100 Hz) was the AguaClara standard 
•Some gravity flow water supplies don’t  (2011‐2015)
have much elevation difference 
between source and storage tank • Summer 2015 new designs have head loss of approximately 
40 cm
•Velocity gradient (G)  • Expect smaller flocs (but still captured by plate settlers)
•Higher allows lower residence time • Less sedimentation of flocs in flocculator
•Higher  ̅ results in smaller flocs 
• Smaller flocculator
• Casey Garland has tested  ̅ values as high as 340 Hz

CEE 4540: Sustainable Municipal Drinking Water Treatment

Monroe Weber-Shirk 3
 1/10/2017

Our current choice of parameter 
The design inputs for flocculation
that sets energy input is head loss
•We need collisions and thus G •Head loss is independent of temperature
is a logical design specification •Velocity gradient is f(temperature)
•We need to specify energy use •Start with ( ̅ ,  ̅ ) and coldest temperature
Option 1

•Velocity gradient ‐ ̅ Higher G means •Calculate 

smaller flocs and more •Calculate hFloc
•Energy dissipation rate ‐ ̅ elevation drop (head
•Start with (hFloc, 
Option 2 ̅ ) and coldest temperature
•Total head loss ‐ loss) through
•Calculate  ̅
flocculator Current

•Or t () More time helps

approach
•Calculate 
diffusion of coagulant • ̅ (and hence  ̅ ) will increase when the 
Current approach
nanoparticles to clay flocculator is operated at warmer 
surfaces
temperatures due to decrease in viscosity

Design the reactor geometry to  Solve for channel width to set
get the target velocity gradient
Kinetic energy dissipated
per residence time
Continuity
S W
Rectangular
geometry
is height of one expansion zone.
Could be the depth of water if the only
expansion is from the 180 degree bend
This is our general equation relating velocity
gradient to reactor geometry
 
constraints on viable solutions
A
This is the minimum channel width if
we set = 3 and set the expansion
height to equal water depth
Elevation view
As channel gets narrower the spacing between baffles
gets larger.
Channels narrower than this would have barely any or
negative baffle overlap!

Minimum number of expansions  Our Design Approach
per depth of flocculator (given W) Given energy ( or  ) and G

Eliminate S
•Start big and then design the details
•Calculate volume of flocculator (AguaClara 
approach as of 

Solve for maximum distance

•Split it into channels summer 2015)

between expansions, , •Then design baffles, and obstacles to fill the 

using =6 channels to get target 

Round up to get the minimum number of

•We can use this design approach because 
we are assuming that we will design for high 
expansions per depth of the flocculator
efficiency (3< ⁄ <6) and thus we don’t 
have to add extra volume to account for 
inefficiencies. (Don’t forget this requirement!)

CEE 4540: Sustainable Municipal Drinking Water Treatment

Monroe Weber-Shirk 4
 1/10/2017

Design Algorithm (as of 2016) Viscous Collision Potential per Flow 
Start with  and G Expansion (the detailed perspective)
1. Velocity gradient and flocculator volume given 
Collision potential for one flow expansion
head loss and collision potential
Height of one expansion zone (in a vertical flow flocculator)
2. Minimum channel width required to achieve 
⁄ >3 and required for constructability Hydraulic residence time for one expansion zone
3. Number of channels by taking the total width and  These are the average velocities through the expanded flow area
dividing by the minimum channel width (floor) Energy dissipation rate is energy loss per time
4. Channel width (total width over number of 
channels) Collision potential is a function of velocity.
5. Maximum distance between expansions This suggests that a flocculator would perform
6. Minimum number of expansions per baffle space poorly if the flow rate were decreases. I don’t
7. Actual distance between expansions know if anyone has ever demonstrated that!
8. Baffle spacing
9. Calculate the obstacle width to obtain the same jet 
expansion conditions as produced by the 180 
degree bend

Almost Real Designs
Velocity guidelines?
(Flocculator exit depth of 2 m)
0.5
• What sets maximum channel width? •Why does V 
• What sets minimum channel width? increase with  0.4

• Why this cycle of channel widths?

Velocity (m/s)
flow rate?
Number of channels 1 2 3 4 5
•Why does V 
0.3
1.2

Sheet width increase in 

Channel width (m)

0.2
1

steps?
0.8

Maximum
•Why does V  0.1
0 50 100

Flow rate (L/s)

150 200

Human hip
0.6
Design remain  Max (10 State)
Minimum Max (Schulz)
0.4 constant  Design
0 50 100 150 200
Min (10 State)
Flow rate (L/s)
above 70 L/s? Min (Schulz)

Design Scaling 
More details
(Design Engine version 7099)
10 L/s 20 L/s 50 L/s 70 L/s
0.53 m wide 0.55 m wide 0.56 m wide 0.72 m wide •The ports between channels should have the 
channels channels channels channels same cross sectional area as WS
4.33 m long 5.90 m long 6.68 m long 7.27 m long •The number of chambers per canal (except in 
the last canal) is even – the number of baffles is 
odd
•The number of chambers in the last canal is odd 
– the number of baffles is even
•Why?

CEE 4540: Sustainable Municipal Drinking Water Treatment

Monroe Weber-Shirk 5
 1/10/2017

extra extra
Use a Pipe with orifices to make a 
Estimate the orifice diameter
flocculator for small flows (S=D)
K eV 3 4Q  DPipe
2

2
DPipe
 G2  V  H e   HS DPipe DOrifice 
 DPipe 2 Continuity K eorifice    1
2He   D2
 vc Orifice 
  vc  
K eorifice  1
Here we assume that S is like D We need to estimate Ke!
Ke  4Q 
3
•The head loss for these orifices spaced so 
 G2    closely may be less than what we calculate
2 HS DPipe   DPipe 2 
•Vena contracta may not be as severe for orifices 
1
Round to nearest inner pipe that are close to the inner diameter of the pipe
 Ke  4Q  
3 7
DPipe  2    diameter? Or round down to •Insufficient length for full expansion before 
2 
 HS  G     get higher velocities to next orifice
prevent sedimentation?

extra extra
Estimate the orifice diameter using  Use a Pipe with orifices to make a 
the correct value of Ke flocculator for small flows (H=D)
V2
H e   HS DPipe he  K e K eV 3 4Q
H e  DPipe
 G2  V 
 DPipe 2 Continuity
2g
2He
Need to find actual Ke given pipe diameter to develop target G
Here we assume that S is like D
ghe
G
 Replace residence time with volume/Q 3
K e  4Q 
4 gheQ G 2 H e DPipe
2
 G2   
2 DPipe   DPipe 2 
G he 
 H e DPipe
2
4 gQ

V 2
16Q 2
G  H e DPipe
2 2
 3 DPipe
7
G 2 HSMax
he  K e  Ke  Ke 
2g 2 g 2 DPipe
4
4 gQ 32Q 3 1
Round to nearest inner pipe
 K  4Q 3  7
DPipe DPipe  e2   diameter? Or round down to
DOrifice   2 G     get higher velocities to
 vc  K eorifice  1  prevent sedimentation?

An interesting design
CEPIS Horizontal Flow Flocculator
No this wasn’t AguaClara…

CEE 4540: Sustainable Municipal Drinking Water Treatment

Monroe Weber-Shirk 6
 1/10/2017

A few Reflections Reflection Questions
•Floc size doesn’t seem to be a significant   max
•How does the collision 
constraint for flocculator design potential in a flocculator 
•We may increase energy dissipation rate  change with flow rate?
significantly as we experiment with maintaining 
small flocs that primary particles can attach to
•What is the ratio of  to  ̅
for well designed hydraulic 
•Our broad goal is to maximize performance at 
minimum cost. Thus cost minimization may be an  flocculators?
important constraint for setting the target  •Why might mechanical 
velocity gradient. flocculators break more flocs 
•Maintaining the flocs in suspension is another  than hydraulic flocculators?
important constraint

Reflection Questions Reflection Questions
•What is the relationship 
•What are some 
0.4S
0.6S

between potential energy 
alternate  loss and the average velocity 
geometries? gradient in a flocculator?
•How else could you  •How did AguaClara get 
generate head loss  around the 45 cm limitation?
to create collisions? •How does the non 
uniformity of  (or G) 
influence efficiency of 
energy use?

Conclusions
•Energy dissipation rate determines the 
spacing of the baffles.
•Energy is used most efficiently to create 
collisions when the energy dissipation rate is 
uniform. Therefore H/S between 3 and 6 is 
best.
•Collision potential is a function of geometry 
and a function of flow rate

CEE 4540: Sustainable Municipal Drinking Water Treatment

Monroe Weber-Shirk 7